Infectious diseases are one of the most intimidating threats, responsible for an immense burden of disabilities and deaths. The fundamental basis of any appropriate treatment is to identify different variants of pathogens through rapid, specific, sensitive, and cost-effective diagnostic tools. The available conventional diagnostic methods are mainly based on immunology, culture and microscopy, and a few molecular biology assays, which are still used as gold standards. However, each methods has their own benefits and limitations in terms of processing time, cost and well-trained personnel. Metagenomic next-generation sequencing (mNGS) allows high-throughput and simultaneous identification of pathogens unbiasedly, especially for pathogens that are difficult to culture, such as M. tuberculosis, viruses, anaerobic bacteria, fungi and novel& pathogens, solving the shortcomings of traditional culture. The emerging approach is rapidly moving to clinical laboratories for pathogen detection of infectious diseases. However, there are some problems when it applies to clinical utility. Contamination is currently one of the biggest bottlenecks of mNGS. Sample collection, nucleic acid extraction, library construction and sequencing, reagents, etc, may introduce contamination. Also, the clinical indications, experimental procedures, quality management, performance verification and report interpretation of the technology need to be more standardized. Besides, digital PCR (dPCR) is a cutting-edge method for absolute quantification of nucleic acid with high accuracy and precision. Compared with conventional real-time quantitative PCR, dPCR has advantages in amplifying target templates at low concentration and without calibration curves. Due to its outstanding performance, dPCR has been widely used in nucleic acid tests of infectious diseases. The key step in dPCR is the generation of a large number of uniform and separate partitions containing target templates by chamber-based or droplet-based strategies. Droplet-based approaches rely on microfluidic methods that split samples into monodisperse droplets, and it can incorporate large numbers of droplets in a single device with high throughput, low cost, and simple operation processes. Several newly developed methods for droplet generation have been reported, such as cross-interface oscillation, cross-interface emulsification, and centrifugal microchannel array. However, the efficiency, reliability, and system-level integration of these methods still requires further studies.